Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

Model Question paper 1

1,. Answer any four of the following questions. 2marks
Define Data Mining. Give an example for Data Mining.

1. Define Data Mining. Give an example for Data Mining.

• Data Mining Definition: Data mining refers to the process of

discovering patterns, trends, and insights from large datasets
using various techniques such as statistical analysis, machine
learning, and artificial intelligence.

• Example: In retail, data mining can be used to analyze

customer purchase patterns, identify associations between
products, and predict future buying behavior.
What are ETL Tools?

1. What are ETL Tools?

• ETL Tools Definition: ETL stands for Extract, Transform, Load.

ETL tools are software applications that facilitate the
extraction of data from various sources, transforming it into a
desired format, and loading it into a target database for
analysis and reporting.
Define Classification and Classification Problem.
1. Define Classification and Classification Problem.

• Classification Definition: Classification is a supervised

learning technique in machine learning that involves
categorizing data into predefined classes or labels based on
its features.

• Classification Problem: A classification problem is a type of

machine learning problem where the goal is to predict the
categorical class labels of new, unseen instances based on the
patterns learned from a training dataset.
What is an Entropy? Give an example.

1. What is Entropy? Give an example.

• Entropy Definition: Entropy is a measure of the impurity or

1 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

• Entropy Definition: Entropy is a measure of the impurity or

disorder in a set of data. In decision trees and information
theory, it is used to quantify the uncertainty associated with a
random variable.

• Example: In a binary classification problem, if a dataset

contains an equal number of instances from two classes, the
entropy is maximized, indicating high disorder or uncertainty.
Define Entropy

1. Define Entropy.

• Entropy Definition: In information theory and machine

learning, entropy is a measure of the average amount of
information or uncertainty associated with a set of data. It is
used to evaluate the impurity or disorder in a dataset.
Define Large Itemsets.

1. Define Large Itemsets.

• Large Itemsets Definition: In association rule mining, large

itemsets refer to sets of items (items that frequently occur
together) that meet a specified support threshold. Support is
the proportion of transactions in a dataset that contain a
particular set of items. Large itemsets are a key concept in
discovering frequent patterns in transactional data.
Part B
II. Answer any four of the following questions.
4x5=20
Explain the features of Data Mining,

• Data Variety: Data mining deals with various types of data,

including structured, unstructured, and semi-structured data from
different sources.

• Data Volume: Data mining handles large volumes of data to

extract meaningful patterns and insights.

• Data Velocity: The speed at which data is generated, processed,

and analyzed, allowing real-time insights.

• Data Veracity: Ensures the reliability and accuracy of the data

used for mining.

2 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

• Data Complexity: Involves dealing with complex relationships and

patterns within the data.

Explain Bayesian Classification with an example.

Naive Bayes classification is a simple but powerful algorithm that is

used for predicting the class or category of an object based on its
attributes. It's called "naive" because it makes a strong assumption
that the attributes are independent of each other, even though this
might not always be true in real-world situations.

How Naive Bayes Classification Works:

1. Training Phase:

• During the training phase, the algorithm analyzes a set of

labeled data, where each object is associated with a known
class.

• It calculates the probability of each class occurring in the data

(prior probability, P(C)).

• It also calculates the probability of each attribute value

occurring for each class (conditional probability, P(X|C)).

2. Prediction Phase:

• When a new, unlabeled object (tuple) needs to be classified,

the algorithm estimates the likelihood of it belonging to each
class.

• It does this by combining the probabilities of each attribute

value for each class, based on what it learned during the
training phase.

• The class with the highest probability is chosen as the

predicted class for the new object.

Steps in Simple Terms:

1. Learn from Training Data:

• Look at a set of examples where you know the class of each

object.

3 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

• Count how often each class occurs (prior probability).

• Count how often each attribute value occurs for each class
(conditional probability).

2. Predict for New Object:

• For a new object with certain attribute values:

◦ Multiply the conditional probabilities of each attribute

value for each class.

◦ Multiply this result by the prior probability of each class.

◦ The class with the highest result is predicted as the class

for the new object.

Example:
Let's say you are classifying emails as either spam (C1) or not spam
(C2). You look at the words in each email as attributes.

• During training, you count how often each word occurs in spam
and non-spam emails.

• You also count how many emails are spam and non-spam.

• During prediction, given a new email, you calculate the likelihood

of it being spam or non-spam based on the words it contains.

• Multiply these likelihoods by the prior probabilities of spam and

non-spam.

• The class with the higher result is your prediction.

Explain K Nearest Neighbors (KNN) Algorithm with an example.

KNN, or k-Nearest Neighbors, is a simple and widely used algorithm

in data mining and machine learning for classification and regression
tasks. It's a non-parametric, instance-based learning algorithm that
makes predictions based on the majority class or average value of its
k-nearest neighbors in the feature space.

Key Concepts:
1. Instance-Based Learning:

• KNN is an instance-based learning algorithm, meaning it

4 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

doesn't explicitly learn a model during training. Instead, it

memorizes the training instances.

2. Distance Metric:

• The algorithm relies on a distance metric (e.g., Euclidean

distance, Manhattan distance) to measure the similarity
between instances in the feature space.

3. Parameter k:

• The parameter k represents the number of nearest neighbors

to consider when making predictions. It's a crucial factor in
the algorithm's performance.

How KNN Works:

1. Training:

• During the training phase, KNN memorizes the feature vectors

and their corresponding class labels (for classification) or
target values (for regression).

2. Prediction:

• Given a new, unseen instance, KNN identifies the k-nearest

neighbors in the feature space based on the chosen distance
metric.

• For classification, it assigns the majority class among the k

neighbors to the new instance.

• For regression, it calculates the average of the target values of

the k neighbors and assigns that as the predicted value.

Steps in Simple Terms:

1. Choose k:

• Decide on the number of neighbors (k) to consider when

making predictions.

2. Calculate Distances:

• Measure the distance between the new instance and all

instances in the training set using a distance metric (e.g.,
Euclidean distance).

5 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

3. Identify Neighbors:

• Identify the k-nearest neighbors with the shortest distances to

the new instance.

4. Majority Vote or Average:

• For classification: Assign the majority class among the k

neighbors to the new instance.

• For regression: Calculate the average of the target values of

the k neighbors.

Example:
Let's say you have a dataset with instances representing different
types of fruits based on features like weight and sweetness. You want
to predict the type of fruit for a new instance.

• Choose k, say k = 3.

• Measure the distances between the new instance and all instances
in the dataset.

• Identify the 3 nearest neighbors.

• If 2 neighbors are apples and 1 is an orange, the prediction is

apple.

Considerations:
• Choice of k: The value of k affects the algorithm's performance.
Too small k may lead to noisy predictions, while too large k may
result in overly smooth predictions.

• Distance Metric: The choice of distance metric depends on the

nature of the data.

• Computationally Intensive: For large datasets, the algorithm can

be computationally intensive during prediction.
What is CART( Classification and Regression Tree)? How it works?

A Classification and Regression Tree (CART) is a decision tree

algorithm used for both classification and regression tasks. It's a
popular and versatile machine learning algorithm that recursively
splits the dataset into subsets based on the most significant attribute,

6 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

splits the dataset into subsets based on the most significant attribute,
resulting in a tree-like structure.

Key Concepts:
1. Decision Tree:

• A tree-like model where each internal node represents a

decision based on the value of a particular attribute.

• Each leaf node represents the outcome or predicted value.

2. Splitting Criteria:

• The algorithm selects the attribute and the split point (or
threshold) that best separates the data into homogeneous
subsets.

• For classification, common criteria include Gini impurity and

entropy.

• For regression, the mean squared error (MSE) is often used.

3. Recursive Splitting:

• The dataset is split recursively until a stopping criterion is met

(e.g., maximum depth, minimum samples per leaf).

• Each split further refines the decision boundaries.

4. Classification:

• For classification tasks, the leaf nodes represent the predicted

class based on majority voting.

5. Regression:

• For regression tasks, the leaf nodes represent the predicted

value based on the average of the target values in that node.

How CART Works:

1. Root Node:

• The algorithm selects the attribute and split point that best
separates the entire dataset.

2. Splitting:

• The dataset is split into subsets based on the chosen attribute

and split point.

7 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

and split point.

• This process is repeated for each subset until a stopping

criterion is met.

3. Leaf Nodes:

• The terminal nodes (leaves) contain the final predictions or

classifications.

4. Prediction/Classification:

• For a new instance, it traverses the tree from the root to a leaf,
making decisions based on attribute values.

• The predicted class or value is determined by the leaf node

reached.

Example:
Let's consider a classification task where we want to predict whether a
passenger survived or not based on features like age, gender, and
ticket class.

• The root node might split the data based on gender.

• The next level might split based on age.

• The leaf nodes might represent different survival outcomes.

For a regression task, the target might be the price of a house based
on features like the number of bedrooms and square footage.

• The tree would split the dataset based on features to create leaves
that represent predicted house prices.

Applications:
• Classification: Predicting outcomes like spam or non-spam emails,
customer churn, etc.

• Regression: Predicting numeric values like house prices,

temperature, etc.

• Interpretability: Decision trees are human-readable and can help

understand the decision-making process.

CART is a powerful algorithm with the ability to handle complex

relationships in data. However, it's prone to overfitting, and

8 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

relationships in data. However, it's prone to overfitting, and

techniques like pruning are often applied to prevent this.
Explain Different Methods for Calculating the Distance between
Clusters.

• Euclidean Distance: Measures the straight-line distance between

two points in Euclidean space.

• Manhattan Distance: Calculates the sum of the absolute

differences between the coordinates of two points.

• Cosine Similarity: Measures the cosine of the angle between two

vectors, often used in text mining.

• Minkowski Distance: Generalizes Euclidean and Manhattan

distances by introducing a parameter (p) that controls the
distance calculation.
What are parallel and distributed algorithms? Explain the categories of
parallel and distributed algorithms.

• Parallel Algorithms: These algorithms perform multiple

operations simultaneously on different processors, enhancing
computational speed. Categories include:

◦ Data Parallelism: Distributes subsets of data across

processors.

◦ Task Parallelism: Distributes different tasks across processors.

• Distributed Algorithms: These algorithms involve multiple

interconnected systems working collaboratively. Categories
include:

◦ Message Passing: Systems communicate by passing

messages.

◦ Shared Memory: Systems access shared memory locations.

◦ Distributed Data: Systems operate on distributed datasets.

Part C
III. Answer any four of the following questions. 4x8 32
How does Data Mining work? Explain The Phases involved in Data
Mining.

1. Data Collection: Gather relevant data from diverse sources.

9 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

2. Data Cleaning: Remove inconsistencies, errors, and irrelevant

information from the dataset.

3. Data Exploration: Understand the data's characteristics,

relationships, and patterns.

4. Feature Selection: Identify relevant features that contribute to the

mining task.

5. Data Transformation: Convert data into a suitable format for

analysis.

6. Modeling: Apply various data mining algorithms to generate

models or patterns.

7. Evaluation: Assess the performance of the models using metrics.

8. Deployment: Implement and integrate the models into business

processes.
Explain the Social Implications of data mining.

• Privacy Concerns: Data mining often involves analyzing personal

information, raising concerns about the privacy and security of
individuals.

• Discrimination: Biases in data or algorithms can result in

discriminatory outcomes, affecting certain groups unfairly.

• Surveillance: Extensive data mining may contribute to increased

surveillance, affecting personal freedom and autonomy.

• Transparency: Lack of transparency in data mining processes can

lead to mistrust and skepticism.
Differentiate ID3, C4.5 and CART Algorithms.

1. ID3 (Iterative Dichotomiser 3):

• Objective: ID3 is designed for constructing decision trees with

the primary goal of classification.

• Attribute Handling: It is capable of handling categorical

attributes efficiently.

• Attribute Selection: ID3 employs a top-down, recursive

approach. At each node, it selects the attribute that maximizes
information gain. Information gain is a measure of the

10 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

reduction in uncertainty about the class labels after a split.

• Entropy: The algorithm uses entropy as a metric to quantify

the uncertainty or impurity in a dataset. Information gain is
calculated based on the reduction of entropy after a split.

2. C4.5:

• Objective: Developed as an extension of ID3, C4.5 aims to

overcome some of ID3's limitations and extends its
applicability.

• Attribute Handling: C4.5 can handle both categorical and

continuous attributes, making it more versatile.

• Attribute Selection: Instead of information gain, C4.5

introduces the gain ratio to address the bias towards
attributes with more levels. Gain ratio normalizes information
gain by taking into account the intrinsic information of the
attribute.

• Pruning: C4.5 employs a pruning mechanism to avoid

overfitting by post-pruning the decision tree after it is built.
This helps in generalizing the model to new, unseen data.

3. CART (Classification and Regression Tree):

• Objective: Developed by Leo Breiman, CART is a versatile

algorithm suitable for both classification and regression tasks.

• Tree Structure: Unlike ID3 and C4.5, CART constructs binary

trees, where each node has two children.

• Attribute Selection (Classification): For classification tasks,

CART uses the Gini index as a measure of impurity. The
algorithm selects the attribute and split that minimizes the
weighted Gini index, aiming to create pure nodes.

• Attribute Selection (Regression): For regression tasks, CART

uses variance reduction. It selects the attribute and split that
minimizes the variance within the resulting nodes.

• Cost-Complexity Pruning: CART incorporates cost-complexity

pruning, a method to control the size of the tree. Pruning is
done by iteratively removing branches that do not
significantly contribute to the model's predictive accuracy.

11 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

significantly contribute to the model's predictive accuracy.

Explain different Attribute Selection Measures (ASM) used in
Classification.

1. Information Gain (IG):

• Definition: Information Gain is a metric used in decision tree

algorithms to quantify the effectiveness of an attribute in
reducing uncertainty about the class labels.

• Calculation: It is calculated by taking the entropy (or Gini

impurity) of the parent node and subtracting the weighted
sum of entropies (or Gini impurities) of the child nodes after a
split.

• Use in Decision Trees: In a decision tree, the attribute with the

highest Information Gain is chosen as the splitting attribute at
each node during the tree-building process.

2. Gain Ratio:

• Definition: Gain Ratio is an enhancement of Information Gain

introduced by the C4.5 algorithm to address its bias towards
attributes with more levels.

• Calculation: Gain Ratio is calculated by dividing the

Information Gain by the intrinsic information of the attribute.
Intrinsic information is a measure of the amount of
information contained in an attribute regardless of the class
labels.

• Use in Decision Trees: Gain Ratio is used to adjust Information

Gain, providing a normalized measure that helps prevent
overfitting to attributes with a large number of levels.

3. Gini Index:

• Definition: The Gini Index is a measure of impurity used in

decision tree algorithms, particularly in CART (Classification
and Regression Tree).

• Calculation: It is calculated by summing the squared

probabilities of each class in a dataset subtracted from one.
Lower Gini Index values indicate a more pure dataset.

• Use in Decision Trees: In CART, the attribute and split that

result in the lowest Gini Index are chosen at each node during

12 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

result in the lowest Gini Index are chosen at each node during
the tree-building process. The goal is to create pure nodes
where most instances belong to a single class.

4. Chi-Square:

• Definition: Chi-Square (χ²) is a statistical test that measures

the independence between two categorical variables, such as
an attribute and the class label in the context of decision
trees.

• Calculation: In decision trees, Chi-Square compares the

observed distribution of class labels after a split with the
expected distribution if the attribute and class label were
independent.

• Use in Decision Trees: Chi-Square is often employed as a

criterion for splitting categorical attributes. If the Chi-Square
test indicates that the attribute and class label are
independent, the split is considered less useful.
Explain How Partitional Minimum Spanning Tree Algorithm works?
Write Partitional Minimum Spanning Tree Algorithm.

Partitional Minimum Spanning Tree Algorithm:

1. Input: A dataset with points in a multidimensional space.

2. Distance Calculation: Compute the distance between all pairs of

points.

3. Minimum Spanning Tree (MST): Construct the minimum

spanning tree using Kruskal's or Prim's algorithm.

4. Partitioning: Divide the MST into subtrees.

5. Cluster Formation: Each subtree represents a cluster.

13 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

Write Apriori Algorithm and explain with an example.

• Apriori Algorithm:

1. Input: A database of transactions, each containing a set of

items.

2. Itemset Generation (k=1): Identify frequent individual items.

3. Candidate Generation: Create candidate itemsets of length

k+1 from frequent itemsets of length k.

4. Support Counting: Count the support (frequency) of each

candidate itemset.

5. Pruning: Remove candidate itemsets below a specified

support threshold.

6. Repeat for k = k+1: Iteratively generate larger itemsets until

no new frequent itemsets are found.

14 of 29 06-02-2024, 20:12
Model Question paper 1 https://awaisahmed.notion.site/Model-Question-paper-1-0de375d19c08...

1. Example: Consider a retail dataset with transactions. Apriori

can identify frequent itemsets like {milk}, {bread}, {milk, bread},
etc. Pruning eliminates itemsets with support below a
threshold, leaving only those deemed frequent.

15 of 29 06-02-2024, 20:12